An Integrated Stochastic Framework for Modeling Bank Financial Risks under Macroeconomic and Geopolitical Instability

Abstract

For Ukrainian banks, the years 2021-2025 were marked not only by wartime shocks but also by changes in the way separate financial risks interacted. This article studies these interactions for the TOP-20 banks by regulatory capital, using official statistics of the National Bank of Ukraine. The proposed integrated risk indicator Ltot combines four risk contours: credit, market, liquidity and operational/buffer-related. Its construction is based on an economic logic rather than a purely technical aggregation. Credit risk is treated as the main loss-generating contour; foreign exchange-market and liquidity risks are interpreted as channels of shock transmission; and the operational/buffer-related component reflects the ability of a bank either to absorb or to amplify these shocks. The full MS-DCC-copula-Bayesian architecture is used as a conceptual framework, since the available Ukrainian data form a short panel and do not allow reliable direct estimation of the complete specification. The empirical validation therefore relies on a more transparent set of tools: VaR, Expected Shortfall, panel VAR(2), EWMA-based dependence estimates, FEVD, CUSUM diagnostics and stress simulation.The results show that the fall in integrated risk after the initial wartime shock should not be interpreted as the disappearance of vulnerabilities. In 2025, Ltot increased again, mainly because of renewed credit and portfolio risks. Expected Shortfall was highest in 2021-2022, while the combined stress scenario for 2025 produced the strongest tail-risk effect. The VAR and FEVD results are especially important for the logic of the model: at the third forecast horizon, liquidity shocks explain up to 20.27% of the variance of the credit-risk contour. This means that liquidity risk is not only a supporting indicator but also a channel through which credit vulnerability may develop. The findings suggest that a simple additive treatment of bank risks may understate their interaction under wartime and structural instability.